Before you start: set your sound

• Use headphones or speakers — this study cannot be done without sound.
• Set your volume to a comfortable level now, and keep it the same throughout the study. Changing the volume part-way makes your ratings inconsistent.
• Make sure your device is not muted (check the silent switch on phones and tablets).
• We will ask you to confirm that the sound and video are working once before you begin and once during the study. These checks take a few seconds each.

What you will do

• Each page shows 4 videos of the same speech, produced by different systems.
• Watch each video to the end (with sound on). The rating buttons unlock afterwards.
• Rate every video independently on 3 aspects, using a 1–5 scale.
• There are … pages of ratings in total.
• Please do not reload or use the browser back button.

What to rate

• Naturalness — How natural does the gesture motion look, in terms of resembling real human movement?
• Smoothness — How smooth is the motion — is it free from abrupt stops, jitter, or unnatural jerks?
• Speech-Gesture Synchrony — How well do the gestures align with the rhythm and timing of the speech?

About your data

• We record only your Prolific ID, your ratings, and how long each page took. No names, free text, or recordings of you are collected.
• The data is used for academic research on gesture generation and is reported only in aggregate. Your Prolific ID is removed after the analysis is complete.
• The data is stored on a server in the EU (France) and analysed by researchers at Nagoya Institute of Technology, Japan.
• You may stop at any time by returning the study on Prolific.
• Questions: k.kameyama.311@stn.nitech.ac.jp